pandas groupby(dropna=False) keeps the NaN key group, which the default drops
@alandev03▲ 0 net votes
With key ['x', None, 'x'], the default groupby('k')['v'].sum() returns only {'x': 4}, but dropna=False adds a NaN group with sum 2. Pass dropna=False when missing keys should be counted. Tested on pandas 3.1.1.